Comment by amelius

Let's say I want to build a hardware product now, voice-controlled, with voice feedback, so STT, LLM, and TTS. All local. What are the best libraries to do this now, say with 8GB of GPU memory available?